Machine Learning model deployment with Flask framework

Used Cars Price Prediction Application

Objective:

1. To build a Machine learning regression model to predict the selling price of the used cars based on the different input features like fuel_type, kms_driven, type of transmission etc.
2. Deploy the machine learning model with the help of the flask framework.

Dataset Information:

Dataset Source: https://www.kaggle.com/datasets/nehalbirla/vehicle-dataset-from-cardekho?select=CAR+DETAILS+FROM+CAR+DEKHO.csv

This dataset contains information about used cars listed on www.cardekho.com

• Car_Name: Name of the car
• Year: Year of Purchase
• Selling Price (target): Selling price of the car in lakhs
• Present Price: Present price of the car in lakhs
• Kms_Driven: kilometers driven
• Fuel_Type: Petrol/diesel/CNG
• Seller_Type: Dealer or Indiviual
• Transmission: Manual or Automatic
• Owner: first, second or third owner

1. Import required libraries

## importing libraries 
%matplotlib inline
import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
import seaborn as sns
import warnings
warnings.filterwarnings('ignore')

import sklearn
from sklearn.model_selection import train_test_split
from sklearn.preprocessing import StandardScaler
from sklearn.preprocessing import LabelEncoder

from sklearn.metrics import accuracy_score, r2_score
from sklearn.ensemble import RandomForestRegressor, RandomForestClassifier

import pickle

2. Load the dataset

## Reading the dataset 'loan_approval_data.csv'
try:
    df = pd.read_csv('car+Data.csv')
except FileNotFoundError: # if file path is invalid of the file does not exist
    print("File not found.")
except pd.errors.EmptyDataError: # if the file is empty or has no data
    print("No data")
except pd.errors.ParserError: # if the file has malformed or invalid data that cannot be parsed.
    print("Parse error")
except UnicodeDecodeError: # if the file has a different encoding than the default or specified one.
    print("Encoding error")
except Exception: # handling other exceptions
    print("Error encountered! Please review your code")

##Fetching first few rows to verify the dataset has loaded
df.head()

	Car_Name	Year	Selling_Price	Present_Price	Kms_Driven	Fuel_Type	Seller_Type	Transmission	Owner
0	ritz	2014	3.35	5.59	27000	Petrol	Dealer	Manual	0
1	sx4	2013	4.75	9.54	43000	Diesel	Dealer	Manual	0
2	ciaz	2017	7.25	9.85	6900	Petrol	Dealer	Manual	0
3	wagon r	2011	2.85	4.15	5200	Petrol	Dealer	Manual	0
4	swift	2014	4.60	6.87	42450	Diesel	Dealer	Manual	0

3. Check the shape and basic information of the dataset.

## Checking the shape of dataset
df.shape

(301, 9)

• 301 rows
• 9 columns

## Information of whole dataset
df.info()

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 301 entries, 0 to 300
Data columns (total 9 columns):
 #   Column         Non-Null Count  Dtype  
---  ------         --------------  -----  
 0   Car_Name       301 non-null    object 
 1   Year           301 non-null    int64  
 2   Selling_Price  301 non-null    float64
 3   Present_Price  301 non-null    float64
 4   Kms_Driven     301 non-null    int64  
 5   Fuel_Type      301 non-null    object 
 6   Seller_Type    301 non-null    object 
 7   Transmission   301 non-null    object 
 8   Owner          301 non-null    int64  
dtypes: float64(2), int64(3), object(4)
memory usage: 21.3+ KB

Observations:

• There are 301 rows and 9 columns
• The columns include various data types such as integer, float and object
• There is no missing data.
• The DataFrame contains information related Used Cars Price Prediction Application, including attributes like Car_Name, Year, Selling_Price, Present_Price, Kms_Driven, and various skill-related attributes.
• Some columns like 'Car_Name', 'Fuel_Type','Seller_Type','Transmission' is in string format and required conversion to numeric column for analysis.

4. Check for the presence of the duplicate records in the dataset? If present drop them

## the number of duplicated records in the dataset
df.duplicated().sum()

2

• Two records are duplicated

# Identifying duplicated records
duplicate_records = df[df.duplicated()]

# Checking if there are any duplicate records
if not duplicate_records.empty:
    # If duplicates are found, inform the user and proceed to drop them
    print("Duplicate records found. Dropping duplicates...")
    
    # Dropping duplicate records
    df = df.drop_duplicates(keep='first')
    
    # Informing the user that duplicates have been dropped and providing the updated shape of the data
    print("Duplicates dropped. Updated shape of the data:", df.shape)
else:
    # If no duplicate records are found, inform the user
    print("No duplicate records found.")

Duplicate records found. Dropping duplicates...
Duplicates dropped. Updated shape of the data: (299, 9)

5. Drop the columns which you think redundant for the analysis.

## dropping the redundant columns owner
df.drop('Car_Name', axis=1,inplace=True)

6. Extract a new feature called 'age_of_the_car' from the feature 'year' and drop the feature year

##lets check what is the 'year' column
df['Year']

0      2014
1      2013
2      2017
3      2011
4      2014
       ... 
296    2016
297    2015
298    2009
299    2017
300    2016
Name: Year, Length: 299, dtype: int64

## Calculating age using 'Year' :: current-year - given-year and Dropping the 'Year' column
from datetime import date

df['age_of_the_car'] = date.today().year - df['Year']
df.drop('Year',axis=1,inplace=True)

df.head()

	Selling_Price	Present_Price	Kms_Driven	Fuel_Type	Seller_Type	Transmission	Owner	age_of_the_car
0	3.35	5.59	27000	Petrol	Dealer	Manual	0	10
1	4.75	9.54	43000	Diesel	Dealer	Manual	0	11
2	7.25	9.85	6900	Petrol	Dealer	Manual	0	7
3	2.85	4.15	5200	Petrol	Dealer	Manual	0	13
4	4.60	6.87	42450	Diesel	Dealer	Manual	0	10

7. Encode the categorical columns

## Checking categorical column in dataframe
cat_columns = df.select_dtypes('O')
cat_columns.head(3)

	Fuel_Type	Seller_Type	Transmission
0	Petrol	Dealer	Manual
1	Diesel	Dealer	Manual
2	Petrol	Dealer	Manual

## Printing the unique value counts of each records
print(df['Fuel_Type'].unique())
print(df['Seller_Type'].unique())
print(df['Transmission'].unique())

['Petrol' 'Diesel' 'CNG']
['Dealer' 'Individual']
['Manual' 'Automatic']

## perform label encoding for 'Object' type columns
le = LabelEncoder()

for i in cat_columns:
    df[i] = le.fit_transform(df[i])
    
df.head()

	Selling_Price	Present_Price	Kms_Driven	Fuel_Type	Seller_Type	Transmission	Owner	age_of_the_car
0	3.35	5.59	27000	2	0	1	0	10
1	4.75	9.54	43000	1	0	1	0	11
2	7.25	9.85	6900	2	0	1	0	7
3	2.85	4.15	5200	2	0	1	0	13
4	4.60	6.87	42450	1	0	1	0	10

8. Separate the target and independent features.

## storing traget column 'Selling_Price' in variable Y and other independant features in variable X
X = df.drop('Selling_Price', axis=1)
y = df['Selling_Price']

print('Independant Features : ')
X.head()

Independant Features : 

	Present_Price	Kms_Driven	Fuel_Type	Seller_Type	Transmission	Owner	age_of_the_car
0	5.59	27000	2	0	1	0	10
1	9.54	43000	1	0	1	0	11
2	9.85	6900	2	0	1	0	7
3	4.15	5200	2	0	1	0	13
4	6.87	42450	1	0	1	0	10

print('Target Feature : ')
y.head()

Target Feature : 

0    3.35
1    4.75
2    7.25
3    2.85
4    4.60
Name: Selling_Price, dtype: float64

9. Split the data into train and test.

## splitting the data
X_train, X_test, y_train,y_test = train_test_split(X,y, test_size=0.3, random_state=0)
print(X_train.shape, X_test.shape)
print(y_train.shape, y_test.shape)

(209, 7) (90, 7)
(209,) (90,)

10. Build a Random forest Regressor model and check the r2-score for train and test.

# Create a Random Forest Regressor
rf_regressor = RandomForestRegressor()

# Train the classifier
rf_regressor.fit(X_train,y_train)

# Prediction on test data
y_train_pred = rf_regressor.predict(X_train)
y_test_pred = rf_regressor.predict(X_test)

# Accuracy test for both train and test data
r2_train = r2_score(y_train, y_train_pred)
r2_test = r2_score(y_test, y_test_pred)

# Printing the r2 score both test and train 
print('r2_score train:',r2_train)
print('r2_score test',r2_test)

r2_score train: 0.9813695797063068
r2_score test 0.9019381048341197

11. Create a pickle file with an extension as .pkl

#saving model to disk
pickle.dump(rf_regressor,open('model.pkl','wb'))

# Loading the pickle file
model = pickle.load(open('model.pkl','rb'))

12. Create new folder/new project in visual studio/pycharm that should contain the "model.pkl" file make sure you are using a virutal environment and install required packages.

a) Create a basic HTML form for the frontend

Create a file index.html in the templates folder and copy the following code.

Document

Used Car Price Predictor

{{ prediction_text }}

Enter Car Details:

Age of the car(In years)

Present Showroom Price(In lakhs)

Kilometers Driven

Owner Type (0/1/3)

Fuel type

Seller Type

Transmission type

b) Create app.py file and write the predict function

Copy the code below and paste it in 'app.py' in vs code.

from flask import Flask,render_template, request, jsonify import pickle import numpy as np import sklearn app = Flask(__name__) # open and load the pickle file provided in read mode. model = pickle.load(open('model.pkl','rb')) @app.route('/', methods = ['GET']) def Home(): return render_template('index.html') @app.route('/predict', methods = ['POST']) def predict(): if request.method == 'POST': Present_Price = float(request.form['Present_Price']) Kms_Driven = int(request.form['Kms_Driven']) Owner = int(request.form['Owner']) Fuel_Type = int(request.form['Fuel_Type']) Age_of_the_car = int(request.form['Age_of_the_car']) Seller_Type = int(request.form['Seller_Type']) Transmission = int(request.form['Transmission']) # Predict function to read the values from the UI and predict the price value. prediction = model.predict([[Present_Price,Kms_Driven,Owner,Fuel_Type,Age_of_the_car,Seller_Type,Transmission]]) output = round(prediction[0],2) return render_template('index.html', prediction_text="You can sell your car at {} lakhs".format(output)) if __name__ = "__main__": app.run(debug=True)

13. Run the app.py python file which will render to index html page then enter the input values and get the prediction.

Run the app.py python file Using VS Code or CMD Prompt(Terminal)

Step 1: CMD Prompt

• run python app.py

Step 2: Run the Cell

• Run the notebook cell containing the Flask application. This will start the Flask server within the notebook.

Step 5: Interact with the Form

• Below the cell, you'll see a link like http://127.0.0.1:5000/. Click on it to open the form in your web browser. You can now enter input values, submit the form, and get the prediction displayed.

Attaching Screenshots of the prediction value acheived after running the program.

1. Run the app.py python file

from IPython.display import Image
Image(filename='ss_recordings\Run_file_in_vs_code.png')

2. render to index html page

Image(filename='ss_recordings\Link_given_after_running_in_vs_code.png')

3. Enter the input values

Image(filename='ss_recordings\enter_the_input_values.png')

Image(filename='ss_recordings\enter_the_input_values (2).png')

4. Get the prediction

Image(filename='ss_recordings\getting_the_prediction.png')

5. Working demo

• The recordings is available in the file directory 'ss_recordings'

Happy Learning :)